01
The chat and agentic loop:
Focuses on GitHub.
AI & Machine Learning / 13:37
Run local agentic AI on the Mac using MLX
AI & Machine Learningmacos
Key Points
- Run AI agents locally with privacy, low latency, and offline access.
- Covers how MLX advancements and Mac hardware make powerful agentic workflows possible entirely on-device.
- Covers local coding agents, Xcode integration, multi-Mac scaling, and tool integration with MLX.
- The session moves through The chat and agentic loop, Local agentic AI stack, Setting up their own agent, Making agents fast, Concurrency and distributed inference, More examples.
- Key concepts include MLX, OpenCode, GitHub, HuggingFace, OpenAI, HTTP, Metal, DeepSeek.
- Platform coverage: macos.
Condensed Flow
02
Local agentic AI stack:
Focuses on HuggingFace, MLX.
03
Concurrency and distributed inference:
Focuses on MLX.
04
More examples:
Focuses on MLX.
More Detail
Additional details
- Detailed flow: The chat and agentic loop -> Local agentic AI stack -> Setting up their own agent -> Making agents fast -> Concurrency and distributed inference -> More examples.
- APIs and concepts to recognize: MLX, OpenCode, GitHub, HuggingFace, OpenAI, HTTP, Metal, DeepSeek, SwiftUI, macOS 26.2, RDMA.
- Version and support notes focus on MLX, DeepSeek, macOS 26.2, RDMA.
- Implementation focus: MLX, OpenCode, GitHub, HuggingFace, DeepSeek, OpenAI, HTTP.